UBC Entity Discovery and Linking & Diagnostic Entity Linking
نویسندگان
چکیده
This paper describe the runs submitted by the UBC team at TAC-KBP 2014 for both English Entity Discovery and Linking (EDL) and Diagnostic Entity Linking (DEL) tasks. Our main interest was to compare the performance between two totally different name entity recognizer systems and to combine them with three different name entity disambiguation systems that were developed for the TACKBP 2013 EL task. Therefore, we tried 6 possible detection-disambiguation combinations for EDL task. The results show that all system combinations attain similar scores, and that the best result is obtained by combining a supervised name entity recognizer with a random forest classifier for disambiguation. For the DEL task our best performance was obtained by disambiguating mentions with a Personalized PageRank algorithm. All systems reported to both tracks reached at least top 10, and also, scored between best and median performance in all the cases.
منابع مشابه
The Effect of Transitive Closure on the Calibration of Logistic Regression for Entity Resolution
This paper describes a series of experiments in using logistic regression machine learning as a method for entity resolution. From these experiments the authors concluded that when a supervised ML algorithm is trained to classify a pair of entity references as linked or not linked pair, the evaluation of the model’s performance should take into account the transitive closure of its pairwise lin...
متن کاملEstimating the Parameters for Linking Unstandardized References with the Matrix Comparator
This paper discusses recent research on methods for estimating configuration parameters for the Matrix Comparator used for linking unstandardized or heterogeneously standardized references. The matrix comparator computes the aggregate similarity between the tokens (words) in a pair of references. The two most critical parameters for the matrix comparator for obtaining the best linking results a...
متن کاملStanford-UBC Entity Linking at TAC-KBP
This paper describes the joint Stanford-UBC knowledge base population system for the entity linking task. We modified our 2009 approach, which was based on frequencies of Wikipedia back-links, providing a context-independent mapping from strings to possible Wikipedia titles. We then built on this foundation, taking into account named-entity recognition (NER) and coreference resolution informati...
متن کاملUBC Entity Linking at TAC-KBP 2013: random forests for high accuracy
This paper describe our systems and different runs submitted for the Entity Linking task at TAC-KBP 2013. We developed two systems, one is a generative entity linking model and the other is a supervised system reusing the scores of the previous model using random forests. Our main research interest is Named Entity Disambiguation task and we thus performed a very naive clustering of NIL instance...
متن کاملStanford-UBC at TAC-KBP
This paper describes the joint Stanford-UBC knowledge base population system. We developed several entity linking systems based on frequencies of backlinks, training on contexts of anchors, overlap of context with the text of the entity in Wikipedia, and both heuristic and supervised combinations. Our combined systems performed better than the individual components, which situates our runs bett...
متن کامل